An automation control platform running state monitoring method and system

Abstract

The application relates to the technical field of data processing, and discloses an automatic control platform operation state monitoring method and system. The method comprises the following steps: a terminal captures and processes platform multi-source data, a network transmission engine performs encrypted transmission, a decoding system analyzes and constructs a digital twin model, a double-flow fusion network outputs a health score, the health score is mapped to a fault atlas to predict risks, an early warning system generates an intervention strategy according to the risks and executes the intervention strategy, and a monitoring control closed loop is established. The application realizes comprehensive, safe and accurate monitoring of the operation state of the automatic control platform, and provides an effective prediction and intervention mechanism before a fault occurs.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to an automatic control platform running state monitoring method and system. BACKGROUND

[0002] At present, the automatic control platform is widely used in industrial production, energy management, transportation and other fields, and its safe and stable operation is crucial to ensure production continuity and equipment safety. The existing automatic control platform running state monitoring technology mainly includes simple monitoring method based on threshold judgment, abnormal detection technology based on statistical model and fault diagnosis method based on expert system. The monitoring method based on threshold judgment sets the normal range of key parameters, and triggers an alarm when the parameters exceed the threshold; the abnormal detection technology based on statistical model uses historical data to construct a parameter distribution model to identify abnormal points deviating from the normal distribution; the fault diagnosis method based on expert system relies on a pre-prepared rule base to infer possible fault causes according to symptom combinations. In addition, some advanced systems also use single-dimensional machine learning algorithms for state evaluation, such as support vector machines, decision trees and neural networks, to improve the accuracy of fault identification.

[0003] However, the existing technology has many shortcomings such as single data collection, insufficient transmission security, rough feature extraction, one-sided state evaluation and inaccurate early warning. First, most systems only collect single type of data (such as parameter data), lack comprehensive analysis of video stream and control instruction stream, resulting in insufficient state monitoring dimension; second, the security protection measures in the data transmission process are simple, which is difficult to cope with security threats in complex network environment; third, the feature extraction mostly uses static method, which cannot effectively capture the dynamic running characteristics of the system; fourth, the state evaluation is usually based on a single algorithm model, lacking fusion analysis of static attributes and dynamic behaviors, reducing the comprehensiveness of the evaluation; finally, the early warning system is mostly triggered by fixed threshold, which cannot dynamically adjust the response strategy according to the running environment and task importance, causing high false alarm rate and inaccurate intervention measures. These shortcomings seriously limit the effect of automatic control platform running state monitoring, and cannot meet the strict requirements of system reliability and predictive maintenance in current high complexity industrial environment. SUMMARY

[0004] The present application provides an automatic control platform running state monitoring method and system for realizing comprehensive, safe and accurate monitoring of the running state of the automatic control platform, and providing effective prediction and intervention mechanism before failure occurs.

[0005] In a first aspect, the application provides an automatic control platform running state monitoring method, which comprises: capturing platform video stream, control instruction stream and parameter stream through a collection terminal, applying a lossless compression algorithm to the video stream, performing time sequence marking on the control instruction stream, and performing noise suppression on the parameter stream to generate a multi-source heterogeneous data packet; inputting the multi-source heterogeneous data packet into a network transmission engine, which adjusts a transmission strategy and applies an encryption protocol according to network quality to form an encrypted data stream; using a decoding system to perform analysis on the encrypted data stream, importing the analysis result into a feature extraction module, and combining a knowledge base and a data mining technology to construct a digital twin feature model; applying a double-flow fusion network to the digital twin feature model, integrating through an attention mechanism after processing in a static feature channel and a dynamic feature channel, and outputting a health score matrix; mapping the health score matrix to a fault knowledge graph, calculating abnormal similarity through an inference engine, combining a long short-term memory network to predict a fault risk, and generating a prediction report; triggering an early warning system based on the prediction report, generating an intervention strategy according to a risk level and issuing an execution instruction through a secure channel to establish a monitoring and control link.

[0006] In a second aspect, the application provides an automatic control platform running state monitoring system, which comprises:

[0007] a marking module for capturing platform video stream, control instruction stream and parameter stream through a collection terminal, applying a lossless compression algorithm to the video stream, performing time sequence marking on the control instruction stream, and performing noise suppression on the parameter stream to generate a multi-source heterogeneous data packet;

[0008] an encryption module for inputting the multi-source heterogeneous data packet into a network transmission engine, which adjusts a transmission strategy and applies an encryption protocol according to network quality to form an encrypted data stream;

[0009] an importing module for using a decoding system to perform analysis on the encrypted data stream, importing the analysis result into a feature extraction module, and combining a knowledge base and a data mining technology to construct a digital twin feature model;

[0010] an integrating module for applying a double-flow fusion network to the digital twin feature model, integrating through an attention mechanism after processing in a static feature channel and a dynamic feature channel, and outputting a health score matrix;

[0011] a prediction module for mapping the health score matrix to a fault knowledge graph, calculating abnormal similarity through an inference engine, combining a long short-term memory network to predict a fault risk, and generating a prediction report;

[0012] An establishing module is configured to trigger an early warning system based on the prediction report, the early warning system generates an intervention strategy according to a risk level and issues an execution instruction through a secure channel to establish a monitoring and control link.

[0013] The third aspect of the present application provides a computer device, comprising: a memory and at least one processor, the memory has instructions stored therein; the at least one processor invokes the instructions in the memory to enable the computer device to execute the automatic control platform running state monitoring method.

[0014] The fourth aspect of the present application provides a computer readable storage medium, the computer readable storage medium has instructions stored therein, when the instructions are executed on a computer, the computer executes the automatic control platform running state monitoring method.

[0015] In the technical scheme provided in the application, the multi-port acquisition terminal simultaneously captures platform video stream, control instruction stream and parameter stream, realizes multi-dimensional and all-around data acquisition, provides rich original data basis for comprehensive monitoring of platform state, simultaneously applies targeted processing algorithms to different types of data, such as lossless compression algorithm to video stream, time sequence marking to control instruction stream, and noise suppression to parameter stream, ensures data quality and transmission efficiency, and significantly reduces data storage and transmission burden; the network transmission engine dynamically adjusts transmission strategy according to a network quality evaluation matrix and applies multi-layer encryption protocol, effectively improves safety and reliability of data transmission in complex network environment, and guarantees integrity and confidentiality of monitoring data; the decoding system successfully combines the digital twin feature model constructed by combining a knowledge base and data mining technology with a feature extraction module, successfully fuses static structure information and dynamic behavior features of the platform, and provides accurate feature expression for comprehensive evaluation of platform state; the double-stream fusion network fully utilizes stability of static features and time sequence of dynamic features through parallel processing of static feature channels and dynamic feature channels and attention mechanism integration, and a health score matrix generated by the double-stream fusion network can reflect the health state of the platform from multiple angles; the health score matrix is mapped to a fault knowledge graph, and combined with a long short-term memory network to perform abnormal similarity calculation and fault risk prediction, which significantly improves the accuracy and foresight of fault prediction, so that the system can identify potential risks before the actual occurrence of faults; the early warning system can adaptively adjust response measures according to risk levels and current environment of the platform based on a hierarchical early warning and intervention strategy generation mechanism triggered by the prediction report, realizes the change from passive response to active intervention, and significantly reduces the failure rate and influence range. The artificial intelligence algorithms and models applied in the scheme, such as interframe lossless compression algorithm, wavelet transform filtering technology, double-stream fusion network, attention mechanism, long short-term memory network and fault knowledge graph, not only fully consider the characteristics and needs of data in the industrial control field, but also more specifically solve the problems of low data processing efficiency, incomplete feature extraction, one-sided state evaluation and inaccurate early warning in traditional monitoring methods, realize the intelligentization of the whole process from data acquisition, transmission processing, feature extraction, state evaluation to early warning intervention, so that the system can maintain high-precision monitoring effect and prediction ability in complex industrial environments, and significantly improve the reliability, safety and maintainability of the automatic control platform. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical scheme of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are some embodiments of the application, and those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0017] Figure 1An embodiment of the automatic control platform running state monitoring method in the present application is shown in the figure;

[0018] Figure 2 An embodiment of the automatic control platform running state monitoring system in the present application is shown in the figure;

[0019] Figure 3 An embodiment of the computer device in the present application is shown in the figure. DETAILED DESCRIPTION

[0020] The present application provides an automatic control platform running state monitoring method and system. The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above-mentioned figures are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" or "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to only those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0021] For ease of understanding, the specific process of the embodiments of the present application is described below. Please refer to Figure 1 An embodiment of the automatic control platform running state monitoring method in the present application includes:

[0022] Step S101, capture platform video stream, control instruction stream and parameter stream through acquisition terminal, apply lossless compression algorithm to video stream, perform time sequence marking to control instruction stream, and perform noise suppression to parameter stream, to generate multi-source heterogeneous data packets;

[0023] Step S102, input the multi-source heterogeneous data packets into a network transmission engine, and the network transmission engine adjusts the transmission strategy according to the network quality and applies an encryption protocol to form an encrypted data stream;

[0024] Step S103, use a decoding system to perform analysis on the encrypted data stream, and import the analysis result into a feature extraction module, and the feature extraction module combines a knowledge base and a data mining technology to construct a digital twin feature model;

[0025] Step S104, apply a double-flow fusion network to the digital twin feature model, and the double-flow fusion network integrates through an attention mechanism after processing in a static feature channel and a dynamic feature channel, to output a health score matrix;

[0026] Step S105, mapping the health score matrix to the fault knowledge graph, calculating the anomaly similarity through the reasoning engine and combining the long short-term memory network to predict the fault risk, and generating a prediction report;

[0027] Step S106, triggering the early warning system based on the prediction report, the early warning system generating an intervention strategy according to the risk level and issuing an execution instruction through a secure channel to establish a monitoring and control link.

[0028] It can be understood that the execution subject of the present application can be an automatic control platform running state monitoring system, and can also be a terminal or a server, and the specific place is not limited. The server is taken as an example for description in the embodiments of the present application.

[0029] Specifically, first, the terminal simultaneously connects the video output interface, control interface and communication interface of the automation control platform to capture the platform video stream, control instruction stream and parameter stream. Specifically, the multi-port acquisition terminal synchronously acquires the original video signal, control instruction signal and running parameter signal. For the original video signal, a hardware-accelerated lossless inter-frame compression algorithm is used for processing. The algorithm compares consecutive frames of images, identifies and retains key region information, and generates compressed video data. For example, when only a local area of the automation control platform display interface changes, the compression algorithm only records the pixel information and position coordinates of the changed area, thereby greatly reducing the data volume. For the control instruction signal, the acquisition terminal uses a high-precision clock source for microsecond-level time sequence labeling, records the transmission time and response time difference of each instruction, constructs a complete instruction time sequence chain, and forms control instruction data with time stamps. This enables the system to accurately track the execution status and delay of the control instruction. At the same time, wavelet transform filtering technology is applied to the running parameter signal to eliminate environmental interference and electromagnetic noise, extract key running parameters of the platform, and generate noise-reduced parameter data. Finally, the compressed video data, control instruction data with time stamps and noise-reduced parameter data are synchronized and calibrated according to a unified time reference, and data format header information and check codes are added to encapsulate multi-source heterogeneous data packets with internal correlation. After the multi-source heterogeneous data packets are input into the network transmission engine, they are first divided into three priority levels of control data packets, monitoring data packets and state data packets according to the importance of the data. The network state detector continuously collects the current network bandwidth occupancy, communication delay value and data packet loss rate to construct a network quality evaluation matrix. The network transmission engine allocates transmission resources for data packets of different priority levels according to the matrix to ensure that control data packets are transmitted preferentially under network bandwidth constraints. Subsequently, the multi-source heterogeneous data packets are protected by a multi-layer protection structure. First, the symmetric encryption algorithm (such as AES-256) is used to process the data content, then the security protocol protection (such as TLS 1.3) is applied to the transmission channel, and finally the secure tunnel (such as IPSec) is established at the network layer to form an encrypted data stream. Digital signature information and time stamp data are also embedded in the encrypted data stream to realize error detection and automatic data recovery functions during transmission through the segment verification mechanism.

[0030] After receiving the encrypted data stream, the decoding system performs layered decryption through the multi-layer decryption processor, sequentially removes the network layer security tunnel, the transport layer security protocol and the data encryption layer, and restores the original structure of the data. For the decrypted video stream data, the image stream analysis engine is used for frame sequence reconstruction; for the control instruction stream and parameter stream data, the protocol identifier is used for data separation and format restoration, and finally the original monitoring data is obtained. The feature extraction module extracts a standard operating parameter set from the original monitoring data based on the platform type knowledge graph, including operating status identification, performance values and environmental variables. The adaptive time window analyzer performs time series slicing on the original monitoring data, calculates the parameter fluctuation amplitude and change rate, and identifies platform-specific operating characteristics. Finally, the standard operating parameter set and the platform-specific operating characteristics are combined by weight to construct a digital twin feature model with spatial and temporal attributes, which includes static feature channels and dynamic feature channels. When processing the digital twin feature model, the dual-flow fusion network first uses the feature separator to split the model into static feature data and dynamic feature data, which are input into the corresponding feature channels. The spatial correlation extractor in the static feature channel performs multi-layer deep analysis on the static feature data, mines the potential relationship between features, and generates a spatial dimension feature vector. The time series pattern recognizer in the dynamic feature channel performs continuity analysis on the dynamic feature data, extracts the parameter change law, and forms a time dimension feature vector. The context perception calculation unit calculates the weight coefficients according to the spatial dimension and time dimension feature vectors based on the current operating environment of the platform. The feature mapping converter projects the weighted feature vectors into the same calculation space and performs data fusion through a nonlinear transformation equation. The platform overall health index, core component state score, key link stability value and abnormal risk warning coefficient are extracted from the fused data to form a multi-dimensional health score matrix.

[0031] The health score matrix is projected to the feature space of the fault knowledge graph through the feature mapper, establishing the coordinate correspondence between the current state and the fault mode. The graph traverser performs bidirectional traversal along the concept association path in the fault knowledge graph, calculates the Euclidean distance values between the current state feature point and each type of fault node, and obtains the abnormal similarity ranking table. The multi-scale analyzer divides the health score historical data into time periods, extracts the hour-level fluctuation features, day-level trend features, and week-level change features, respectively, and forms a multi-scale time series feature set. The feature set is input into the long short-term memory network (LSTM), which saves historical state information through the triple structure of the forget gate, input gate, and output gate, and performs time series extrapolation on the key indicators to generate a future state prediction curve. Combined with the abnormal similarity ranking table and the future state prediction curve, the Bayesian inference calculator determines the most likely fault type, fault probability value, and predicted occurrence time point, and generates a structured fault prediction report. The early warning system extracts the fault type identifier, risk level value, and time urgency indicator from the prediction report, triggers the hierarchical early warning processing procedure. The environmental perception unit collects the current task importance parameters, operating environment condition data, and historical intervention response records of the platform, and constructs a context information library. According to the context information library, the early warning threshold is dynamically adjusted to form a hierarchical early warning threshold value that matches the current scenario. From the intervention strategy library, the set of processing schemes with the highest historical effect score is extracted, and the parameter adjustment amount, load control amplitude, or system switching time is finely calculated through the gradient strategy optimization algorithm to generate a targeted intervention strategy. The intervention strategy is converted into a standard control instruction set, added with a security check code and an execution priority marker, and sent to the on-site control terminal through an encrypted transmission channel. The platform response information during the intervention process is collected and analyzed in real time, and the intervention effect evaluation data and system recovery indicators are recorded to update the intervention effect library and threshold parameter table, completing the closed-loop construction of the monitoring and control link.

[0032] Taking the automation control platform of a chemical plant as an example, the terminal captures the operation interface video stream, control valve instruction stream and temperature and pressure parameter stream from the control platform. After applying the inter-frame lossless compression algorithm to the 1920x1080 resolution video stream, only the changed area data on the interface is recorded, reducing the data volume from about 6MB to 0.8MB per second. The microsecond-level time sequence label is executed on the control valve instruction stream, recording the time difference from the issuance of the instruction to the valve response, and it is found that the response time of a certain key valve gradually increases from the standard 15ms to 28ms. After applying the wavelet transform filtering technology to the temperature and pressure parameter stream, the temperature fluctuation noise of ±0.5℃ is eliminated, and the 0.8℃ abnormal rising trend of the temperature of the reaction kettle within 30 minutes is accurately captured. According to the current 80% bandwidth occupancy and 15ms communication delay, the network transmission engine prioritizes the transmission of control data packets while reducing the sampling frequency of monitoring data packets. The health score matrix output by the double-flow fusion network processing digital twin feature model shows that the stability value of the temperature control loop of a certain reaction kettle decreases from the normal 92 to 78, and is mapped to the "temperature control system fault" node by the fault knowledge graph, with an abnormal similarity of 89%. The long short-term memory network predicts that the fault will develop into a serious fault within 8 hours. The early warning system immediately generates an intervention strategy, adjusts the parameters of the standby temperature control loop and sends a switching instruction, successfully avoiding the possible production line shutdown accident.

[0033] In the embodiments of the present application, the multi-port acquisition terminal simultaneously captures platform video stream, control instruction stream and parameter stream, realizing multi-dimensional and all-round data acquisition, providing rich raw data basis for comprehensive monitoring of platform state, and applying targeted processing algorithms to different types of data, such as lossless compression algorithm for video stream, time sequence marking for control instruction stream, and noise suppression for parameter stream, to ensure data quality and transmission efficiency, significantly reducing data storage and transmission burden; the network transmission engine dynamically adjusts transmission strategy according to the network quality evaluation matrix and applies multi-layer encryption protocol, effectively improving the security and reliability of data transmission in complex network environment, ensuring the integrity and confidentiality of monitoring data; the decoding system combines the digital twin feature model constructed by the knowledge base and data mining technology with the feature extraction module, successfully fusing the static structure information and dynamic behavior characteristics of the platform, providing accurate feature expression for comprehensive evaluation of platform state; the dual-stream fusion network integrates through parallel processing of static feature channel and dynamic feature channel and attention mechanism, fully utilizing the stability of static features and the time sequence of dynamic features, and the generated health score matrix can reflect the health status of the platform from multiple angles; the health score matrix is mapped to the fault knowledge graph, and combined with the long short-term memory network for abnormal similarity calculation and fault risk prediction, significantly improving the accuracy and foresight of fault prediction, enabling the system to identify potential risks before the actual occurrence of faults; the early warning system can adaptively adjust the response measures according to the risk level and the current environment of the platform based on the hierarchical warning and intervention strategy generation mechanism triggered by the prediction report, realizing the transformation from passive response to active intervention, and significantly reducing the failure rate and impact range. The artificial intelligence algorithms and models applied in the present scheme, such as inter-frame lossless compression algorithm, wavelet transform filtering technology, dual-stream fusion network, attention mechanism, long short-term memory network and fault knowledge graph, not only fully consider the characteristics and needs of data in the industrial control field, but also more specifically solve the problems of low data processing efficiency, incomplete feature extraction, one-sided state evaluation and inaccurate early warning in traditional monitoring methods, realizing the intelligentization of the whole process from data acquisition, transmission processing, feature extraction, state evaluation to early warning intervention, enabling the system to maintain high-precision monitoring effect and prediction ability in complex and variable industrial environments, significantly improving the reliability, security and maintainability of the automation control platform.

[0034] In a specific embodiment, the process of step S101 can specifically include the following steps:

[0035] (1) connecting the video output interface, control interface and communication interface of the automation control platform through the multi-port acquisition terminal simultaneously, acquiring raw video signals, control instruction signals and running parameter signals;

[0036] (2) For the original video signal, the inter-frame lossless compression algorithm based on hardware acceleration is used to compare each frame of image, retain the key area information of the image, and generate compressed video data;

[0037] (3) For the control instruction signal, the high-precision clock source is used to mark the microsecond-level time sequence, record the difference value of instruction sending time and response time, construct the complete instruction time sequence chain, and form the control instruction data with timestamp;

[0038] (4) For the running parameter signal, the wavelet transform filtering technology is combined to eliminate environmental interference and electromagnetic noise, extract the key running parameters of the platform, and generate noise reduction parameter data;

[0039] (5) The compressed video data, control instruction data with timestamp and noise reduction parameter data are synchronized and calibrated according to the unified time reference, the data format header information and check code are added, and the multi-source heterogeneous data packet with internal correlation is packaged.

[0040] Specifically, the acquisition terminal connects the VGA, HDMI or DVI video output interface of the platform through the video acquisition card to obtain the original video signal; connects the RS-485, Modbus or Profibus control interface of the platform through the industrial communication conversion module to capture the control instruction signal; and connects the Ethernet, serial port or field bus communication interface of the platform through the data acquisition unit to obtain the running parameter signal. This multi-port parallel acquisition method ensures the integrity and synchronization of the data source, laying a foundation for subsequent comprehensive analysis. For the original video signal collected, a hardware acceleration-based inter-frame lossless compression algorithm is used for processing. The algorithm first compares two consecutive frames of images at the pixel level and calculates the difference image matrix. Then, the boundary of the difference area is identified through region segmentation technology, and the key area with significant changes is extracted. For these key areas, the algorithm retains the complete pixel information; and for the unchanged areas, only the region position index is stored. For example, when monitoring an industrial control panel, if only the status of an indicator light changes from green to red, the algorithm only needs to record the new pixel value and position information of the indicator light area, without repeatedly storing the data of the entire interface. This method reduces the data volume to 10%-30% of the original video stream while ensuring the quality of the video, generating efficient compressed video data.

[0041] For the processing of the control instruction signal, the acquisition terminal uses the built-in high-precision clock source to mark the microsecond-level time sequence. When the control instruction is detected, the timestamp is recorded; when the corresponding response signal is captured, the timestamp is recorded; and the time difference value , as the instruction execution delay value. This process marks every instruction issued by the control platform, such as valve switching instructions, motor start-stop instructions, temperature setting change instructions, etc., forming a complete instruction timing chain. Each instruction data contains instruction type identifier, instruction content, sending timestamp, response timestamp and delay value fields. This time-stamped control instruction data can accurately reflect the response characteristics and instruction execution efficiency of the control system, providing key basis for fault prediction. For running parameter signals, the terminal application uses wavelet transform filtering technology for processing. First, the original parameter signal is decomposed into different frequency components through wavelet decomposition. By analyzing the energy distribution characteristics of each frequency domain, the high-frequency components representing environmental interference and electromagnetic noise are identified. Then, a suitable threshold function is designed to suppress or eliminate these high-frequency components. Finally, the processed frequency components are reconstructed through wavelet reconstruction to obtain the parameter signal after noise elimination. For example, the data collected by the temperature sensor may contain high-frequency noise caused by power fluctuations, which can be effectively filtered out through wavelet transform, retaining the true temperature change trend. This technology can extract key platform running parameters such as temperature, pressure, flow, voltage, current, etc. from the noisy environment, generating high-quality noise-reduced parameter data.

[0042] The acquisition terminal synchronously calibrates the compressed video data, the time-stamped control instruction data and the noise reduction parameter data according to a unified time reference, and eliminates the time deviation caused by different acquisition paths. Then, a unified format header information is added to each type of data, including data type identification, source identification, time stamp, data length and other fields. Next, the check code of the data content is calculated, such as CRC-32 or MD5 check value, to ensure data integrity. Finally, these processed data and their meta information are organized and encapsulated according to the pre-defined data structure specification, forming a multi-source heterogeneous data packet with internal correlation. This data packet structure design enables the subsequent analysis system to accurately identify and process different types of data while maintaining their time synchronization relationship and logical association. Taking the distillation column control system of a certain petrochemical plant as an example, the multi-port acquisition terminal simultaneously connects the HDMI video output of the operator station, the Modbus control interface of the field control system and the Ethernet interface of the temperature sensor network. The original video signal is a 1920×1080 resolution, 30 frames per second full-color picture. After processing by the inter-frame lossless compression algorithm, it is found that only the temperature value display area and the alarm status indication area in the operation interface change frequently, and the rest are basically static, so the algorithm only retains the real-time image data of these key areas, reducing the video data volume from the original 178 MB / minute to 27 MB / minute. When the control instruction signal is time-stamped, it is recorded that the temperature regulating valve instruction usually takes 235 milliseconds from issuance to execution completion, but in a certain operation, this value suddenly increases to 412 milliseconds, and the instruction time sequence chain clearly records this abnormal situation. After filtering the operating parameter signal by wavelet transform, the random fluctuations of ±0.8℃ and the periodic noise caused by 50Hz power supply interference are successfully eliminated, and the 1.2℃ slow downward trend of the distillation column top temperature in 15 minutes is accurately captured. Finally, the three types of processed data are encapsulated into unified multi-source heterogeneous data packets, each containing 5 seconds of synchronized data and SHA-256 check value, ensuring data integrity and correlation during transmission.

[0043] In a specific embodiment, the process of performing step S102 can specifically include the following steps:

[0044] (1) The multi-source heterogeneous data packets are divided into three priority levels of control data packets, monitoring data packets and state data packets according to the importance of the data;

[0045] (2) The network state detector is used to collect the current network bandwidth occupancy rate, communication delay value and data packet loss rate in real time, and construct a network quality evaluation matrix;

[0046] (3) According to the network quality evaluation matrix, transmission resources are allocated to data packets of different priority levels to ensure that control data packets are transmitted preferentially under network bandwidth limited conditions;

[0047] (4) The symmetric encryption algorithm is used to process the data content of the multi-source heterogeneous data packet, then the security protocol is applied to protect the transmission channel, and finally the secure tunnel is established at the network layer to form the multi-layer protection structure of the encrypted data stream;

[0048] (5) The digital signature information and timestamp data are embedded in the encrypted data stream to realize the error detection and automatic data recovery function in the transmission process through the segmented verification mechanism.

[0049] Specifically, after receiving the multi-source heterogeneous data packet, the network transmission engine first divides the data according to the importance of the data. The control data packet contains the instruction content that has a direct control effect on the automation platform, such as device start-stop instruction, parameter adjustment instruction and emergency treatment instruction, etc., which is given the highest priority; the monitoring data packet contains the real-time monitoring data of the key performance parameters and running state information of the platform, such as temperature, pressure, flow and device load, etc., which is given the medium priority; the state data packet contains auxiliary information and environmental parameters, such as historical trend data, environmental humidity and non-critical component state, etc., which is given the lowest priority. This three-level division based on the importance of the data ensures that the core control function is not affected under the condition of limited network resources. The network state detector continuously monitors the current network environment quality, and adopts a distributed detection point strategy. The bandwidth occupancy rate is obtained by calculating the percentage of data flow in available bandwidth within a fixed time window, such as detecting that the current bandwidth occupancy rate is 75%; the communication delay value is obtained by sending a probe packet and calculating the round-trip time, such as measuring that the average delay is 18 milliseconds; the data packet loss rate is calculated by continuously sending test packets with consecutive sequence numbers and calculating the proportion of missing packets at the receiving end, such as measuring that the loss rate is 0.8%. The network state detector organizes these indicators into a network quality evaluation matrix, and the matrix elements contain the historical change trend in the time dimension and the status of different network segments in the space dimension, forming a comprehensive evaluation of the network environment.

[0050] According to the real-time data of the network quality evaluation matrix, the network transmission engine dynamically adjusts the resource allocation strategy. When the network quality is good, all three types of data packets are transmitted at the original sampling rate; when the network bandwidth occupancy rate exceeds 80% or the communication delay value increases by more than 30%, the transmission frequency of monitoring data packets and status data packets is automatically reduced, such as reducing the sampling rate of monitoring data packets from 10 times per second to 5 times per second and extending the update interval of status data packets from 1 second to 5 seconds; when the network is severely congested, the transmission of status data packets is suspended and the monitoring data packets are reduced to the minimum frequency, ensuring the real-time performance and reliability of control data packets. The network transmission engine also implements an adaptive congestion control algorithm that dynamically adjusts the retransmission timeout and transmission window size based on the timeliness requirements of data packets of different priorities, ensuring that the end-to-end delay of control data packets does not exceed the preset threshold. The security protection of data content uses a multi-level encryption architecture. First, the AES-256 symmetric encryption algorithm is used to process the data content itself, and the encryption key is updated regularly through the key management center. The encryption process processes data blocks in segments, with each data block being 128 bits in size, and uses the CBC (Cipher Block Chaining) mode to ensure encryption strength. Subsequently, the TLS 1.3 security protocol is applied to the transmission channel to protect it, establishing an encrypted communication session and implementing identity authentication and key negotiation during data transmission. Finally, an IPSec security tunnel is established at the network layer to encapsulate and encrypt the entire IP packet, preventing eavesdropping and tampering attacks at the network layer. This three-layer protection structure from data content, transmission channel to network layer ensures the confidentiality and integrity of data during transmission.

[0051] To enhance the reliability and tamper-proofing capability of data transmission, additional security information is embedded in the encrypted data stream. Digital signature information is generated based on the asymmetric encryption algorithm RSA-2048, and the private key is signed for the digest of the data content. The receiver can verify the authenticity and integrity of the data through the public key. Timestamp data is provided by a trusted time source, marking the precise time of data generation and transmission, preventing replay attacks and timing chaos. In addition, a segmented verification mechanism is implemented, dividing the data stream into fixed-size data segments, each with Reed-Solomon forward error correction code attached. This allows the receiving end to automatically recover some lost or damaged data when transmission errors are detected. When serious errors that cannot be recovered by error correction are detected, the system triggers a selective retransmission mechanism, requesting only the transmission of the erroneous data segment rather than the entire data packet, improving transmission efficiency and reliability. Taking the actual operation of a factory automation control system as an example, after the network transmission engine receives the multi-source heterogeneous data packet containing the valve control instruction, it identifies it as a control data packet and assigns it the highest priority. At the same time, the network state detector detects that the bandwidth occupancy rate of the current factory intranet reaches 87%, the communication delay rises from the normal 12 milliseconds to 25 milliseconds, and the data packet loss rate is 1.2%. Based on this, the network quality assessment matrix shows that the network is in a moderately congested state. The network transmission engine immediately adjusts the resource allocation strategy, reducing the transmission frequency of status data packets by 50%, maintaining the basic transmission rate of monitoring data packets, and allocating a priority transmission channel for control data packets to ensure that valve control instructions can be delivered to the execution terminal in a timely and reliable manner. In terms of data security processing, the system first uses the AES-256 key valid on the current day to encrypt the valve control instruction, then transmits it through the TLS 1.3 secure channel, and establishes an IPSec tunnel between the core network nodes. These three layers of encryption ensure that even if the data is intercepted, the original control instruction content cannot be parsed. At the same time, the system adds a digital signature based on the factory PKI system and a timestamp accurate to the millisecond to the encrypted data, and applies Reed-Solomon encoding to achieve 8-bit data recoverable forward error correction capability, successfully resisting 3-bit data errors during network jitter, ensuring the integrity and reliability of the control instruction.

[0052] In a specific embodiment, the process of performing step S103 can specifically include the following steps:

[0053] (1) The encrypted data stream is decrypted by a multi-layer decryption processor to remove the encryption protection layer and restore the original data structure;

[0054] (2) For the decrypted video stream data, the image stream analysis engine is used for frame sequence reconstruction, and for the control instruction stream and parameter stream data, the protocol identifier is used for data separation and format restoration to obtain the original monitoring data;

[0055] (3) From the original monitoring data, based on the platform type knowledge graph, the standard operating parameter set is extracted, including operating state identification, performance value and environmental variable;

[0056] (4) The original monitoring data is time-sliced by the adaptive time window analyzer, the parameter fluctuation amplitude and change rate are calculated, and the platform-specific operating characteristics are identified;

[0057] (5) The standard operating parameter set and platform-specific operating characteristics are combined by weight matching, and a digital twin feature model with spatial and temporal attributes is constructed, including static feature channels and dynamic feature channels.

[0058] Specifically, the received encrypted data stream is processed by a multi-layer decryption processor. The decryption process follows the reverse order of encryption. First, the IPSec security tunnel of the network layer is processed, and the tunnel encapsulation is removed after identity verification through pre-shared keys or digital certificates. Then, the TLS 1.3 security protocol of the transport layer is processed, and the session key is used to decrypt the transmission channel content. Finally, the symmetric key decryption algorithm (such as AES-256) is used to decrypt the data content itself. During the decryption process, the decryption processor also verifies the digital signature and timestamp information of the data to ensure the authenticity and timeliness of the data source. After multi-layer decryption, the encryption protection layer is completely removed, and the data is restored to the original structure of the multi-source heterogeneous data packet, retaining complete metadata such as data type identification, data format header information and checksum. For different types of data streams after decryption, the system uses a special parsing engine for processing. For video stream data, the image stream parsing engine first identifies the video frame boundary in the data packet, and then decodes according to the inverse process of the lossless compression algorithm between frames. The static area is reconstructed using position index and reference frame data, and the changing area directly uses new pixel data. The image stream parsing engine also performs frame sequence synchronization and buffer management to handle possible frame loss or out-of-order situations, ensuring that the reconstructed video stream is smooth and coherent. For control instruction stream and parameter stream data, the protocol identifier identifies different communication protocols such as Modbus, Profibus, OPC UA, etc. according to the data format header information, and then applies the corresponding protocol parsing rules to separate and restore the data format. The protocol identifier extracts the type, content, execution status and timestamp of each instruction, as well as the name, value, unit and quality mark of various parameters. Through these professional parsing processes, the system successfully obtains the original monitoring data, laying the foundation for subsequent analysis.

[0059] The feature extraction module first extracts a set of standard operating parameters from the raw monitoring data based on the platform type knowledge graph. The platform type knowledge graph is a structured knowledge base that contains information such as standard operating parameter definitions, normal value ranges, and inter-parameter relationships for different types of automated control platforms. The system queries the knowledge graph for a matching parameter template based on the current monitoring platform's model and configuration information, and then extracts the corresponding operating status indicators (such as operating mode, start-stop status, alarm level, etc.), performance values (such as throughput, response time, processing capacity, etc.), and environmental variables (such as temperature, humidity, power voltage, etc.) from the raw monitoring data. These standard operating parameters form the core indicator set for evaluating the basic operating status of the platform, with clear physical meanings and evaluation standards. The adaptive time window analyzer performs time series processing on the raw monitoring data to identify platform-specific operating characteristics. This analyzer can automatically adjust the analysis window size based on data characteristics, using smaller time windows (such as 100 milliseconds) for high-frequency changing parameters and larger time windows (such as 10 minutes) for low-frequency changing parameters. Within each time window, statistical features such as mean, standard deviation, maximum, and minimum are calculated for the parameters, and dynamic features such as fluctuation amplitude (difference between maximum and minimum), change rate (first-order difference value), acceleration (second-order difference value), and periodicity (identified through fast Fourier transform) are also calculated. These time series features can reflect the operating status change patterns of the platform at different time scales, revealing dynamic behavior patterns that cannot be directly expressed by the standard parameter set.

[0060] The system combines the standard operating parameter set and the platform-specific operating characteristics with weight allocation to construct a digital twin characteristic model. The weight allocation process is based on expert experience rules and historical data analysis results, and assigns importance coefficients to different types of characteristics. Standard operating parameters usually obtain higher basic weights, while platform-specific operating characteristics obtain dynamically adjusted weights according to their relevance to the platform health status. The merged digital twin characteristic model has both spatial and temporal attributes: the spatial attribute reflects the correlation and mutual influence between different parameters, forming a static characteristic channel; the temporal attribute reflects the trend and law of parameter change over time, forming a dynamic characteristic channel. The static characteristic channel focuses on describing the current state of each component of the platform and the relationship between components, while the dynamic characteristic channel focuses on describing the behavior pattern and state transition characteristics of the platform during operation. This dual-channel structure enables the characteristic model to comprehensively capture the operating state information of the platform. Taking the lithography machine control system of a semiconductor manufacturing plant as an example, after the decoding system receives the encrypted data stream, the multi-layer decryption processor successively removes the IPSec tunnel protection, TLS session encryption and AES-256 data encryption, and verifies the effectiveness of the RSA-2048 digital signature, confirming that the data comes from the authorized collection terminal. The image stream analysis engine reconstructs the decrypted video stream, restores the real-time picture of the lithography machine control interface, and clearly displays the exposure parameter setting and alignment state indication. The protocol identifier parses the SECS / GEM protocol data specific to the lithography machine, extracting device control instructions and operating parameters. Based on the lithography equipment knowledge graph, the system extracts a standard operating parameter set, including 25 core indicators such as wafer processing state, exposure energy value, alignment accuracy, chamber pressure, etc. The adaptive time window analyzer uses a 200 millisecond window for the pressure parameter and finds a small fluctuation of ±0.05 Pa; uses a 5 minute window for the temperature parameter and identifies a slow rising trend of 0.3°C / hour; uses a 1 minute window for the alignment accuracy and captures a periodic shift synchronized with the vibration source. The system combines these standard parameters and time sequence characteristics according to the preset weight scheme to construct a digital twin characteristic model of the lithography machine, with its static characteristic channel describing the working state and mutual relationship of each component, and its dynamic characteristic channel recording the change law and device response characteristics of key parameters, providing a comprehensive data basis for subsequent health assessment.

[0061] In a specific embodiment, the process of performing step S104 can specifically include the following steps:

[0062] (1) Use the characteristic separator to split the digital twin characteristic model into static characteristic data and dynamic characteristic data, and input them into the static characteristic channel and the dynamic characteristic channel, respectively;

[0063] (2) The static feature data is analyzed by a multi-layer depth analysis of a spatial correlation extractor in the static feature channel, potential relationships between features are mined, and a spatial dimension feature vector is generated;

[0064] (3) The dynamic feature data is analyzed by a time sequence pattern recognizer in the dynamic feature channel, parameter change rules are extracted, and a time dimension feature vector is formed;

[0065] (4) The spatial dimension feature vector and the time dimension feature vector are calculated by a context perception calculation unit, and the weight coefficients are dynamically changed with the running environment of the platform;

[0066] (5) The weighted spatial dimension feature vector and the time dimension feature vector are projected into the same calculation space by a feature mapping converter, and data fusion is performed through a nonlinear transformation equation;

[0067] (6) The platform overall health index, core component state score, key link stability value and abnormal risk early warning coefficient are extracted from the fused data to form a multi-dimensional health score matrix.

[0068] Specifically, the process of the dual-flow fusion network processing the digital twin feature model is first split by the feature separator. The feature separator adopts a combination of principal component analysis and convolutional neural network to identify and separate static features and dynamic features in the model. Static feature data mainly includes relatively stable information such as platform structure attributes, component relationships, and performance parameter benchmark values, such as controller types, sensor configurations, communication interface parameters, and standard working parameters; dynamic feature data includes time-varying information such as parameter change trends, response modes, and state transition characteristics, such as temperature fluctuation curves, pressure change rates, and instruction response time delay changes. The feature separator calculates the time sensitivity index of the features, and classifies features with sensitivity below the threshold as static features and features with sensitivity above the threshold as dynamic features, thereby completing the binary split of the feature model. The separated static feature data and dynamic feature data are input into the static feature channel and the dynamic feature channel for further processing. The spatial correlation extractor in the static feature channel performs multi-layer deep analysis on the input static feature data. The spatial correlation extractor is composed of a hierarchical neural network, including a feature encoding layer, an association detection layer, and a feature fusion layer. The feature encoding layer uses an autoencoder to reduce the dimension of the original static features and remove redundant information. The association detection layer constructs an association graph between feature nodes through a graph convolution network, where each node represents a feature and the edge weight between nodes represents the association strength between features. The association strength is determined by calculating various indicators such as correlation coefficient, mutual information, and conditional probability between features. The feature fusion layer aggregates information based on the association graph, and strongly associated feature groups are identified as functional modules or subsystems. Through this multi-layer analysis process, the spatial correlation extractor mines the potential relationships between static features, such as control loop composition, system dependency chain, and functional group, and encodes these relationships into fixed-dimensional spatial dimension feature vectors.

[0069] At the same time, the time sequence pattern recognizer in the dynamic feature channel performs continuity analysis on the dynamic feature data. The time sequence pattern recognizer first applies a sliding window technique to the input time sequence data to extract feature segments at different time scales. Then, various time sequence analysis methods such as frequency spectrum analysis, wavelet transform, and recurrent neural network are applied to these feature segments to identify time domain features such as periodic patterns, trend changes, and mutation characteristics. The time sequence pattern recognizer also calculates derivative features such as parameter change rate, fluctuation amplitude, and phase relationship to capture the time sequence dependence and causal relationship between parameters. Finally, the extracted various time sequence features are integrated into a unified format of time dimension feature vector, which encodes the dynamic behavior patterns and state transition characteristics of the platform during operation.

[0070] The context-aware computing unit is responsible for calculating weight coefficients for the spatial dimension feature vector and the time dimension feature vector. This process can be represented by the following mathematical model:

[0071]

[0072]

[0073] wherein, represents the weight coefficient of the i-th spatial dimension feature, represents the weight coefficient of the k-th temporal dimension feature, represents the current environment complexity index, represents the adaptive environment complexity of the i-th spatial dimension feature, represents the adaptive environment complexity of the k-th temporal dimension feature, represents the correlation matrix of the i-th spatial dimension feature and the j-th environmental factor, represents the current value of the j-th environmental factor, represents the sensitivity matrix of the k-th temporal dimension feature and the l-th dynamic condition, represents the current value of the l-th dynamic condition, , , , , and are adjustment parameters. This weight calculation method based on environmental complexity and condition correlation enables the system to dynamically adjust the importance of features according to the current running environment of the platform. The feature mapping converter projects the weighted spatial dimension feature vector and temporal dimension feature vector into the same calculation space and performs data fusion through a nonlinear transformation equation. The projection process uses tensor mapping technology to uniformly convert feature vectors of different dimensions and scales into tensor representation. The fusion process is realized through multi-head attention mechanism and residual connection. The multi-head attention mechanism allows the model to focus on multiple feature combinations simultaneously, capturing the complex interaction between spatial dimension features and temporal dimension features. The residual connection preserves the original feature information, preventing information loss during the fusion process. Through this fusion method, the feature mapping converter generates a comprehensive representation that combines static structural information and dynamic behavior features.

[0074] The platform overall health index is calculated by applying a global scoring function to the fused data, reflecting the overall running condition of the platform. The core component state score is calculated by a local scoring function for key components such as the central processor, control loop and communication unit, reflecting the health status of each key component. The key link stability value is calculated by link response testing for important functional links in the system, such as the instruction transmission link, data processing link and feedback control link, reflecting the stability and reliability of the link. The abnormal risk early warning coefficient is calculated based on pattern matching and trend extrapolation, predicting the potential failure risk of the system. These indicators together form a multi-dimensional health score matrix, comprehensively reflecting the current and future possible health status of the platform.

[0075] Taking the blast furnace automation control platform of a certain steel plant as an example, the feature separator analyzes the digital twin feature model and divides 32 relatively stable features such as controller configuration, sensor distribution and tuyere parameters into static feature data, and 28 time-varying features such as temperature variation trend, pressure fluctuation pattern and flow response characteristics into dynamic feature data. The spatial correlation extractor analyzes the static feature data in multiple layers and finds that there is a strong correlation between the furnace bottom temperature sensor parameters and the cooling water flow controller parameters (correlation coefficient 0.87), identifies the complete "temperature-flow" control loop structure, and excavates the dependency relationship between multiple functional groups, such as the "air volume-fuel-temperature" ternary regulation system. These relationships are encoded into a 96-dimensional spatial dimension feature vector. The time series pattern recognizer analyzes the dynamic feature data and identifies an 8-hour periodic fluctuation pattern of the furnace temperature under normal working conditions, captures a 15-minute time delay relationship between pressure change and raw material addition, and finds a trend feature that the flow regulation response time gradually extends from the standard 3 seconds to 4.5 seconds. These dynamic features are integrated into a 128-dimensional time dimension feature vector. The context perception calculation unit detects that the current blast furnace is in "high yield mode" and the environmental temperature is high (environmental complexity index is 0.78), and accordingly calculates that the average value of the weight coefficient of the spatial dimension feature is 0.65 and the average value of the weight coefficient of the time dimension feature is 0.82, reflecting that the dynamic feature is more important in high temperature and high yield environment. The feature mapping converter fuses and processes the weighted feature vector to generate a comprehensive feature representation. The multi-dimensional health score matrix extracted from the fused data shows that the platform overall health index is 87 (full score 100), the state score of the cooling water control unit among the core components is 75, which is significantly lower than that of other components; the stability value of the "temperature-flow" key link is 0.82, which is in the acceptable range; the abnormal risk early warning coefficient is 0.35, indicating that the system has a moderate risk. These score data indicate that the cooling water control unit needs to be paid attention to.

[0076] In a specific embodiment, the process of performing step S105 can specifically include the following steps:

[0077] (1) Project the multi-dimensional score values in the health score matrix to the feature space of the fault knowledge graph using the feature mapper, and establish the coordinate correspondence between the current state and the fault mode;

[0078] (2) Traverse the fault knowledge graph along the concept association path in both directions through the graph traverser, calculate the Euclidean distance values between the current state feature point and each type of fault node, and obtain an abnormal similarity ranking table;

[0079] (3) Divide the health score historical data into time periods based on the multi-scale analyzer, extract the hour-level fluctuation features, day-level trend features, and week-level change features respectively, and form a multi-scale time series feature set;

[0080] (4) Input the multi-scale time series feature set into the memory unit network, and the memory unit network saves the historical state information through the triple structure of the forgetting gate, the input gate, and the output gate, time series extrapolates the key indicators, and generates a future state prediction curve;

[0081] (5) Combine the abnormal similarity ranking table and the future state prediction curve, and determine the most likely fault type, fault probability value, and predicted occurrence time point through the Bayesian inference calculator;

[0082] (6) Organize the fault type description, risk level score, occurrence probability data, and predicted time information into a structured document, and generate a fault prediction report.

[0083] Specifically, the feature mapper receives the multi-dimensional health score matrix and projects it into the feature space of the fault knowledge graph. The health score matrix contains multi-dimensional score values of platform overall health index, core component state score, key link stability value, and abnormal risk warning coefficient, etc. These indicators constitute the state vector in high-dimensional space. The feature mapper uses dimension reduction mapping algorithm to convert the high-dimensional state vector into coordinate points in the feature space of the fault knowledge graph, preserving the relative relationship between indicators. The fault knowledge graph is a structured knowledge representation, where each fault mode is represented as a region or point in the feature space, and the similarity and causality between faults are represented as the connection between nodes. The feature mapping process establishes the spatial correspondence between the current system state and the known fault modes, enabling the complex multi-dimensional state to be directly compared with fault modes in a unified feature space. The graph traverser performs bidirectional traversal analysis in the fault knowledge graph. Bidirectional traversal includes forward traversal that spreads outward from the current state point and backward traversal that converges inward from each fault node. Forward traversal calculates the Euclidean distance from the current state point to each fault node in the graph, while backward traversal calculates the matching degree between the possible pre-state of each fault node and the current state. The Euclidean distance calculation is based on the square root of the weighted difference square sum of each dimension in the feature space, and the weight is dynamically set according to the importance of the feature to fault identification. The graph traverser also considers the association strength between nodes, preferentially traversing along the path with high association degree, ensuring that the most relevant fault mode can be found. After the traversal process is completed, the system sorts all possible fault modes according to the calculated Euclidean distance values, generating an abnormal similarity ranking table that records the type identifier, similarity score, and key feature difference of each possible fault.

[0084] The multi-scale analyzer performs time period division and feature extraction on the health score historical data. For short-term fluctuations, a sliding analysis is performed using an hourly-level window (usually 1-4 hours) to extract high-frequency features such as fluctuation frequency, amplitude, and mutation characteristics of the parameters; for medium-term trends, a smoothing process is performed using a daily-level window (usually 1-7 days) to extract features such as gradual trend, periodicity, and stability of the parameters; for long-term changes, an aggregation analysis is performed using a weekly-level window (usually 1-4 weeks) to extract low-frequency features such as seasonal changes, aging characteristics, and long-period oscillations of the parameters. Multi-scale analysis uses signal processing techniques such as wavelet transform and empirical mode decomposition to decompose the time series into different frequency components, separating noise and meaningful signals. Through this multi-scale time window analysis, the system can capture both short-term fluctuations and long-term evolution, forming a comprehensive multi-scale time series feature set. The memory unit network analyzes and predicts the multi-scale time series feature set. This network is based on the long short-term memory (LSTM) architecture, containing a triple structure of forget gate, input gate, and output gate. The forget gate determines which historical information needs to be discarded, calculates the forgetting coefficient through the sigmoid function, and filters information that is no longer relevant in the long term; the input gate controls the input proportion of new information at the current time, and through the joint action of the sigmoid function and the tanh function, selectively updates the memory cell state; the output gate determines the final output value according to the current memory cell state and input information. After multiple rounds of training, the memory unit network can effectively capture the complex patterns of health score indicators over time, including long-term dependencies and short-term fluctuation characteristics. The system inputs the processed multi-scale time series feature set into the memory unit network, iteratively calculates the future state of the key health indicators, generates a future state prediction curve, and each point on the curve contains the predicted health indicator value and its confidence interval.

[0085] The Bayesian inference calculator combines the anomaly similarity ranking table and the future state prediction curve for fault prediction analysis. The calculator first establishes the prior distribution of fault probability, setting the initial probability based on the similarity scores in the anomaly similarity ranking table. Then, using the future state prediction curve as evidence, the posterior probability of various fault types is calculated. The Bayesian inference process considers the comprehensive matching degree of historical fault data, current state characteristics and predicted trajectory, and can handle the competition and uncertainty of multiple potential faults. By calculating the probability distribution of each fault type at different time points, the Bayesian inference calculator determines the most likely fault type, the probability value of the fault occurrence and the predicted time point. For fault prediction results with high probability, the system also analyzes the main influencing factors and possible evolution paths leading to the fault. The system organizes the key information of fault prediction into a structured document and generates a fault prediction report. The report contains four main parts: fault type description (detailed description of the possible fault type, impact range and typical symptoms), risk level score (comprehensive risk score calculated based on fault probability, impact range and severity), occurrence probability data (including central probability value and probability distribution interval) and predicted time information (time point and time window when the fault may occur). The report is organized in a standardized XML structure or JSON format for easy system parsing and processing. At the same time, the report also contains visual charts such as risk trend chart, fault probability distribution chart and key parameter prediction curve, which intuitively display the prediction results and provide a basis for subsequent warning and intervention decisions.

[0086] Taking the automatic control system of a chemical plant's reactor as an example, the feature mapper projects the real-time collected health score matrix (containing 92 indicators) into the fault knowledge graph feature space, and locates the coordinate of the current state point. The graph traverser calculates that the Euclidean distance between the state point and the "temperature control loop fault" node is the smallest (0.28), and the distance between the state point and the "pressure sensor offset" node is the second (0.43), generating an abnormal similarity ranking table containing 15 possible faults. The multi-scale analyzer analyzes the health score data of the past 3 months and finds that the temperature control deviation presents a periodic fluctuation of ±2.5°C in a 2-hour window, a gradual increase of 0.8°C / day in a 3-day window, and a step characteristic related to the change of raw material batch in a 2-week window. After processing these multi-scale features, the memory unit network predicts that the temperature control deviation will break through the safety threshold within 72 hours, and the response time of the control loop will increase by 40%. The Bayesian inference calculator integrates these information and determines that "temperature controller servo valve fault" is the most likely fault type (probability 0.82), and predicts that it will reach the critical state affecting production after 68 hours. The system generates a detailed prediction report, including fault type details, risk level (78 / 100), probability distribution curve and predicted time window (68±5 hours), providing sufficient preparation time and clear intervention direction for maintenance personnel.

[0087] In a specific embodiment, the process of performing step S106 can specifically include the following steps:

[0088] (1) Extract the fault type identifier, risk level value and time urgency indicator from the prediction report using the risk resolver, trigger the hierarchical warning processing flow;

[0089] (2) Collect the current task importance parameters, operating environment condition data and historical intervention response records of the platform through the environment perception unit, and build a context information library;

[0090] (3) Dynamically adjust the warning threshold based on the context information library, and form a hierarchical warning threshold value matched with the current scene;

[0091] (4) Extract the set of processing schemes with the highest historical effect score from the intervention strategy library, and perform fine calculation on the parameter adjustment amount, load control amplitude or system switching time through the gradient strategy optimization algorithm, to generate a targeted intervention strategy;

[0092] (5) Convert the intervention strategy into a standard control instruction set, add a security check code and an execution priority marker, and send it to the field control terminal through an encrypted transmission channel;

[0093] (6) Real-time collection and analysis of platform response information during intervention, recording intervention effect evaluation data and system recovery indicators, updating intervention effect library and threshold parameter table, and completing the closed-loop construction of monitoring and control link.

[0094] Specifically, the risk parser, as the front-end processing unit of the early warning system, first extracts three types of key information from the failure prediction report: fault type identification, risk level value, and time urgency index. The fault type identification adopts a hierarchical coding method, including device category, subsystem category, and specific fault mode, such as "CT-03-0542" representing "control system-temperature regulation loop-sensor offset fault". The risk level value is represented based on a standardized scale of 0-100, which quantifies the fault probability value, impact range, and severity in the prediction report, and is divided into three levels: low risk (0-30), medium risk (31-70), and high risk (71-100). The time urgency index reflects the time expectation of fault occurrence, including remaining safe operation time and fault development rate, and is divided into long-term warning (> 72 hours), medium-term warning (24-72 hours), and emergency warning (< 24 hours). The risk parser automatically triggers a hierarchical early warning processing flow that matches the risk level and urgency based on the extracted information, such as high-risk emergency warning that will start the highest-level processing flow, including immediate notification of key personnel and preparation of emergency measures. The environment perception unit collects context information through multiple data sources to build a comprehensive context information library. The task importance parameter is obtained from the production scheduling system, including factors such as the priority, completion degree, planned capacity, and economic value of the current task. The operating environment condition data come from the environment monitoring system, including physical environment factors such as temperature, humidity, power quality, and network status, as well as operation and maintenance environment factors such as personnel allocation, technical support availability, and historical intervention response records extracted from the system log library, including the intervention measures taken in similar situations in the past and their effect evaluation data. The environment perception unit structurally processes these data to generate standardized context information entries, each containing a data item identifier, current value, reference range, and importance weight. These entries are indexed and classified according to device type, fault category, and environmental characteristics to form a context information library that can be quickly queried.

[0095] Based on the context information library, the early warning system dynamically adjusts the early warning threshold. First, the reference threshold set matching the current fault type is extracted from the standard threshold library, including the default trigger conditions of different early warning levels. Then, the sensitivity factor is adjusted by the task importance parameter in the context information, and a lower trigger threshold is used during the execution of important tasks to improve the early warning sensitivity. The fault tolerance factor is adjusted by the environmental condition data, and the judgment standard of normal fluctuation range is appropriately relaxed in harsh environment. The learning factor is adjusted by the historical intervention record, and the current threshold is optimized according to the early warning accuracy in the past similar scenes. This dynamic adjustment mechanism of multiple factors working together ensures that the early warning threshold is highly matched with the current scene, avoiding false alarms caused by excessive sensitivity and missing reports caused by insufficient sensitivity. The final generated hierarchical early warning threshold contains trigger conditions for three levels of light early warning, general early warning and severe early warning, and each level is further subdivided into multiple response strategies according to time urgency. When the early warning is triggered, the intervention decision module retrieves the matching processing scheme from the intervention strategy library. The intervention strategy library is based on historical cases, and each case contains fault feature description, intervention measures taken, intervention result evaluation and expert experience summary. The retrieval process first narrows down the search range based on the fault type identification, and then selects the top N cases with the highest similarity according to the similarity calculation of the current fault characteristics and historical cases as the reference set. The cases in the reference set are sorted according to the historical effect score, and the highest scoring processing scheme set is extracted. These schemes are then refined by the gradient strategy optimization algorithm, and the optimal parameter adjustment amount, load control amplitude or system switching time is calculated for the current actual situation. The optimization algorithm uses simulated annealing strategy, gradually approaches the optimal solution through multiple iterations, and considers the balance between intervention cost and expected effect, finally generates a customized intervention strategy for the current fault.

[0096] The process of converting intervention strategies into standard control instruction sets involves instruction format conversion, security verification, and priority marking. First, abstract intervention strategies are converted into specific control instructions. For example, parameter adjustment strategies are converted into specific parameter setting instructions, and load control strategies are converted into device start-stop or power regulation instructions. Then, a security verification code is added to each instruction, usually generated using the SHA-256 hashing algorithm, to ensure that the instructions are not tampered with during transmission and execution. At the same time, based on fault risk and time urgency, the execution priority of the instructions is marked, and high-priority instructions will be given priority in resource allocation and execution. Finally, the instruction set is sent to the on-site control terminal through an encrypted transmission channel. The encrypted channel usually uses the TLS 1.3 protocol with a two-way authentication mechanism to ensure the security and integrity of the instruction transmission. After receiving the instructions, the on-site control terminal first verifies the security verification code, and then executes the instructions in order of priority, implementing the intervention strategy. Closed-loop monitoring and control are achieved by real-time collection of platform response information during intervention. These information includes the execution status of intervention instructions, the actual value after parameter adjustment, the change of system performance indicators, and the evolution of fault characteristics. Through high-frequency sampling and real-time analysis, intervention effect evaluation indicators such as fault characteristic mitigation rate, system stability recovery speed, and control accuracy are calculated. At the same time, the key indicator change curve during system recovery is recorded, including the transition trajectory from abnormal state to normal state and the stabilization time. These data are used to update two important knowledge bases: the intervention effect library and the threshold parameter table. The intervention effect library records the detailed process and results of each intervention, providing a reference for future similar situations; the threshold parameter table adjusts the early warning threshold and intervention trigger conditions based on the intervention results, optimizing the accuracy of future early warnings. Through this data-driven continuous self-optimization, the monitoring and control link forms a complete closed loop, and the system's fault prediction and intervention capabilities are continuously improved.

[0097] Taking the distribution automation control platform of a certain power system as an example, the risk analyzer extracts the fault type as "PC-07-0124" (distribution circuit-overload protection-trigger delay abnormality) from the prediction report, the risk level is 68 (medium risk), and the time urgency is 36 hours (medium-term warning), which triggers the secondary warning processing flow. The environment perception unit collects that the current task being executed is "urban power supply guarantee" (importance score 92), the environmental temperature reaches 34℃ (exceeding the normal operation reference range), and historical records show that similar faults develop faster under high temperature conditions. Based on these context information, the system dynamically adjusts the standard overload protection trigger delay warning threshold from 25 milliseconds to 18 milliseconds, forming a strict warning threshold for the current high temperature and high load scenario. Retrieve the three sets of processing schemes with the highest historical effect score from the intervention strategy library, calculate the best intervention parameters through the gradient strategy optimization algorithm: adjust the standby line load sharing ratio to 40%, increase the main line overload protection sensitivity by 15%, and arrange maintenance personnel on standby 2 hours in advance. These strategies are converted into 15 specific control instructions, added with SHA-256 security check code and "emergency-3" priority label, and sent to the distribution station control terminal for execution through the TLS encrypted channel. After the intervention is implemented, real-time monitoring shows that the line load is successfully reduced, the overload protection trigger delay is restored from the abnormal 28 milliseconds to the normal 14 milliseconds, and the fault risk score is reduced from 68 to 23. The complete process and effect data of this successful intervention are recorded and used to update the intervention effect library, and the overload protection system warning threshold parameters are optimized, providing valuable experience for subsequent similar situation processing.

[0098] The above describes the automatic control platform operation state monitoring method in the embodiments of the application. The automatic control platform operation state monitoring system in the embodiments of the application is described below. Please refer to Figure 2 An embodiment of the automatic control platform operation state monitoring system in the embodiments of the application includes:

[0099] The marking module is configured to capture platform video streams, control instruction streams, and parameter streams through the acquisition terminal, apply a lossless compression algorithm to the video streams, perform time sequence marking on the control instruction streams, and perform noise suppression on the parameter streams to generate multi-source heterogeneous data packets.

[0100] The encryption module is configured to input the multi-source heterogeneous data packets into a network transmission engine, and the network transmission engine adjusts a transmission strategy and applies an encryption protocol according to network quality to form an encrypted data stream.

[0101] The import module is configured to perform analysis on the encrypted data stream by using a decoding system, and import the analysis result into a feature extraction module. The feature extraction module combines a knowledge base and a data mining technology to construct a digital twin feature model.

[0102] an integration module configured to apply a dual-stream fusion network to the digital twin feature model, the dual-stream fusion network being integrated through an attention mechanism after processing in a static feature channel and a dynamic feature channel, and outputting a health score matrix;

[0103] a prediction module configured to map the health score matrix to a fault knowledge graph, calculate an abnormal similarity through an inference engine, and predict a fault risk in combination with a long short-term memory network to generate a prediction report;

[0104] an establishment module configured to trigger an early warning system based on the prediction report, the early warning system generating an intervention strategy according to a risk level and issuing an execution instruction through a secure channel to establish a monitoring and control link.

[0105] Reference Figure 3 In the embodiments of the present application, a computer device, which can be a server, is also provided, and an internal structure of the computer device can be as shown in Figure 3 The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium. The database of the computer device is configured to store corresponding data in the embodiments. The network interface of the computer device is configured to communicate with an external terminal through a network connection. The computer program is executed by the processor to implement the above method.

[0106] Those skilled in the art can understand that Figure 3 the structure shown in the embodiments of the present application is only a block diagram of part of the structure related to the present application scheme, and does not constitute a limitation on the computer device to which the present application scheme is applied.

[0107] The present application also provides a computer readable storage medium having a computer program stored thereon, and the computer program is executed by the processor to implement the above method. It can be understood that the computer readable storage medium in the embodiments of the present application can be a volatile readable storage medium or a non-volatile readable storage medium.

[0108] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, storage, database or other medium provided by the present application and used in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0109] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-mentioned system, system and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0110] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a number of instructions for making a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various program code storage media.

[0111] The above-described embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for monitoring the operational status of an automated control platform, characterized in that, The method for monitoring the operational status of the automated control platform includes: By capturing platform video stream, control command stream, and parameter stream through the acquisition terminal, a lossless compression algorithm is applied to the video stream, timing marking is performed on the control command stream, and noise suppression is performed on the parameter stream to generate multi-source heterogeneous data packets; The multi-source heterogeneous data packets are input into the network transmission engine, which adjusts the transmission strategy and applies an encryption protocol based on network quality to form an encrypted data stream. The encrypted data stream is parsed using a decoding system, and the parsing results are imported into a feature extraction module. The feature extraction module combines a knowledge base and data mining techniques to construct a digital twin feature model. A dual-stream fusion network is applied to the digital twin feature model. After processing the static and dynamic feature channels, the network integrates them through an attention mechanism to output a health scoring matrix. This process includes: splitting the digital twin feature model into static and dynamic feature data using a feature separator, inputting them into the static and dynamic feature channels respectively; performing multi-layer deep analysis on the static feature data using a spatial correlation extractor within the static feature channel to uncover potential relationships between features and generate spatial feature vectors; performing continuous analysis on the dynamic feature data using a temporal pattern recognizer within the dynamic feature channel to extract parameter variation patterns and form temporal feature vectors; calculating weight coefficients for the spatial and temporal feature vectors based on a context-aware computing unit, with the weight coefficients dynamically changing with the platform's operating environment; projecting the weighted spatial and temporal feature vectors onto the same computational space using a feature mapping converter, and fusing the data through a nonlinear transformation equation; and extracting overall platform health indicators, core component status scores, critical link stability values, and abnormal risk warning coefficients from the fused data to construct a multi-dimensional health scoring matrix. The health score matrix is ​​mapped to the fault knowledge graph, and the anomaly similarity is calculated by the inference engine and combined with the long short-term memory network to predict the fault risk and generate a prediction report. The prediction report triggers an early warning system, which generates an intervention strategy based on the risk level and issues execution instructions through a secure channel, thus establishing a monitoring and control link.

2. The method for monitoring the operating status of an automated control platform according to claim 1, characterized in that, The process involves capturing platform video streams, control command streams, and parameter streams via a data acquisition terminal; applying a lossless compression algorithm to the video streams; performing timing marking on the control command streams; and suppressing noise in the parameter streams to generate multi-source heterogeneous data packets. This includes: The multi-port acquisition terminal is connected to the video output interface, control interface and communication interface of the automation control platform at the same time to acquire raw video signals, control command signals and operating parameter signals. For the original video signal, a hardware-accelerated inter-frame lossless compression algorithm is used to compare each frame of the image, retain key image region information, and generate compressed video data. The control command signal is marked with microsecond-level timing using a high-precision clock source, and the difference between the command transmission time and the response time is recorded to construct a complete command timing chain. The system generates timestamped control command data; for the operating parameter signals, it uses wavelet transform filtering technology to eliminate environmental interference and electromagnetic noise, extracts key platform operating parameters, and generates noise reduction parameter data; the compressed video data, timestamped control command data, and noise reduction parameter data are synchronized and calibrated according to a unified time reference, data format header information and check codes are added, and the data is encapsulated into a multi-source heterogeneous data packet with internal correlation.

3. The method for monitoring the operating status of an automated control platform according to claim 1, characterized in that, The step of inputting the multi-source heterogeneous data packets into the network transmission engine, wherein the network transmission engine adjusts the transmission strategy according to network quality and applies an encryption protocol to form an encrypted data stream, includes: The multi-source heterogeneous data packets are divided into three priority levels according to the importance of the data: control data packets, monitoring data packets, and status data packets. A network quality assessment matrix is ​​constructed by collecting real-time data on current network bandwidth utilization, communication latency, and packet loss rate using a network status detector. Based on the network quality assessment matrix, transmission resources are allocated to data packets of different priority levels to ensure priority transmission of control data packets under network bandwidth constraints. The multi-source heterogeneous data packets are first processed using a symmetric encryption algorithm, then a security protocol is applied to protect the transmission channel, and finally a secure tunnel is established at the network layer, forming a multi-layered protection structure for the encrypted data stream. Digital signature information and timestamp data are embedded in the encrypted data stream, and error detection and automatic data recovery are achieved during transmission through a segmented verification mechanism.

4. The method for monitoring the operating status of an automated control platform according to claim 1, characterized in that, The encrypted data stream is parsed using a decoding system, and the parsing result is imported into a feature extraction module. This feature extraction module combines a knowledge base and data mining techniques to construct a digital twin feature model, including: The encrypted data stream is decrypted layer by layer by a multi-layer decryption processor to remove the encryption protection layer and restore the original data structure; for the decrypted video stream data, the frame sequence is reconstructed using an image stream parsing engine; for the control command stream and parameter stream data, the data is separated and the format is restored using a protocol identifier to obtain the original monitoring data. From the original monitoring data, a standard set of operating parameters is extracted based on the platform type knowledge graph, including operating status identifiers, performance values ​​and environmental variables; The original monitoring data is sliced ​​into time series by an adaptive time window analyzer to calculate the parameter fluctuation amplitude and rate of change, and to identify the platform’s unique operating characteristics. The standard operating parameter set and platform-specific operating features are weighted and merged to construct a digital twin feature model with both spatiotemporal attributes. The digital twin feature model includes static feature channels and dynamic feature channels.

5. The method for monitoring the operating status of an automated control platform according to claim 1, characterized in that, The process of mapping the health score matrix to a fault knowledge graph, calculating anomaly similarity through an inference engine, and predicting fault risk using a long short-term memory network to generate a prediction report includes: The multidimensional score values ​​in the health score matrix are projected onto the feature space of the fault knowledge graph using a feature mapper to establish the coordinate correspondence between the current state and the fault mode. The graph traversaler performs bidirectional traversal along the concept association path in the fault knowledge graph, calculates the Euclidean distance between the current state feature point and various fault nodes, and obtains the anomaly similarity ranking table. Based on the multi-scale analyzer, the historical health score data is divided into time periods, and hourly fluctuation features, daily trend features and weekly change features are extracted to form a multi-scale time series feature set. The multi-scale temporal feature set is input into the memory unit network. The memory unit network saves historical state information through a triple structure of forget gate, input gate and output gate, performs temporal extrapolation on key indicators, and generates future state prediction curves. By combining the anomaly similarity ranking table with the future state prediction curve, the most likely fault type, fault probability value, and expected occurrence time are determined using a Bayesian inference calculator. The fault type description, risk level score, occurrence probability data, and expected time information are organized into a structured document to generate a fault prediction report.

6. The method for monitoring the operating status of an automated control platform according to claim 1, characterized in that, The system triggers an early warning system based on the prediction report. The early warning system generates an intervention strategy according to the risk level and issues execution instructions through a secure channel, establishing a monitoring and control link, including: The risk parser is used to extract the fault type identifier, risk level value and time urgency index from the prediction report to trigger a graded early warning process. The environmental perception unit collects current task importance parameters, operating environment condition data, and historical intervention response records to build a context information database. Based on the context information database, the warning threshold is dynamically adjusted to form a hierarchical warning threshold value that matches the current scenario; Extract the set of treatment schemes with the highest historical effect scores from the intervention strategy library, and use the gradient strategy optimization algorithm to perform fine calculations on parameter adjustment amount, load control range or system switching timing to generate targeted intervention strategies; The intervention strategy is converted into a standard control instruction set, a security check code and an execution priority flag are added, and the instruction is sent to the field control terminal through an encrypted transmission channel. The platform response information during the intervention process is collected and analyzed in real time, the intervention effect evaluation data and system recovery indicators are recorded, the intervention effect library and threshold parameter table are updated, and the closed-loop construction of the monitoring and control link is completed.

7. An automated control platform operation status monitoring system, used to implement the automated control platform operation status monitoring method as described in any one of claims 1 to 6, characterized in that, The automated control platform operation status monitoring system includes: The tagging module is used to capture platform video stream, control command stream and parameter stream through the acquisition terminal, apply lossless compression algorithm to the video stream, perform timing tagging on the control command stream, suppress noise in the parameter stream, and generate multi-source heterogeneous data packets. An encryption module is used to input the multi-source heterogeneous data packets into the network transmission engine, which adjusts the transmission strategy and applies an encryption protocol according to the network quality to form an encrypted data stream. The import module is used to parse the encrypted data stream using the decoding system and import the parsing results into the feature extraction module. The feature extraction module combines a knowledge base and data mining technology to construct a digital twin feature model. An integration module is used to apply a dual-stream fusion network to the digital twin feature model. The dual-stream fusion network processes the static and dynamic feature channels and then integrates them through an attention mechanism to output a health scoring matrix. This includes: using a feature separator to split the digital twin feature model into static and dynamic feature data, which are then input into the static and dynamic feature channels respectively; performing multi-layer deep analysis on the static feature data using a spatial correlation extractor within the static feature channel to uncover potential relationships between features and generate spatial-dimensional feature vectors; performing continuous analysis on the dynamic feature data using a temporal pattern recognizer within the dynamic feature channel to extract parameter variation patterns and form temporal-dimensional feature vectors; calculating weight coefficients for the spatial and temporal feature vectors based on a context-aware computing unit, with the weight coefficients dynamically changing with the platform's operating environment; projecting the weighted spatial and temporal feature vectors onto the same computational space using a feature mapping converter, and performing data fusion through a nonlinear transformation equation; and extracting overall platform health indicators, core component status scores, critical link stability values, and abnormal risk warning coefficients from the fused data to construct a multi-dimensional health scoring matrix. The prediction module is used to map the health score matrix to the fault knowledge graph, calculate the anomaly similarity through the inference engine and combine it with the long short-term memory network to predict the fault risk and generate a prediction report. A module is established to trigger an early warning system based on the prediction report. The early warning system generates an intervention strategy according to the risk level and issues an execution command through a secure channel, thus establishing a monitoring and control link.

8. A computer device, characterized in that, The system includes a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement the automated control platform operation status monitoring method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, the computer program causing a processor, when executed by a processor, to perform the automated control platform operation status monitoring method as described in any one of claims 1 to 6.

Images